EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

refactor: bug fixes and refactor for vlm (#4661)

Browse Source
This commit is contained in:
Mick
2025-03-23 13:48:49 +08:00
committed by GitHub
parent ca75741e86
commit 11577cedb7
31 changed files with 770 additions and 735 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
python/sglang/srt/model_executor/model_runner.py
View File

@@ -273,7 +273,7 @@ class ModelRunner:
if self.model_config.hf_config.architectures == ["DeepseekVL2ForCausalLM"]:
# TODO: deepseek-vl2 does not support radix cache now, set disable_radix_cache=True automatically
logger.info(
"Automatically turn off --chunked-prefill-size and disable radix cache for deekseek-vl2."
"Automatically turn off --chunked-prefill-size and disable radix cache for deepseek-vl2."
)
server_args.chunked_prefill_size = -1
server_args.disable_radix_cache = True